Randall Aiken, cascode vs. pentode, ??

[krash]

in this article:

http://www.blueguitar.org/new/text/thre ... p/6dj8.txt

Randall Aiken is extoling the virtues of using cascode triode input rather than a pentode like EF86. I get the point in generalities but I lack the experience with cascode circuits to understand his admonishment about which circuit to use ("standard" vs. "self-bias") and SRPP and that kind of thing. Can anyone shed some light on this? Maybe point out a schematic that is an example of Aiken's use of cascode input topolgy rather than an EF86?

Thanks-
josh

[jaysg]

I think he saying don't use the first figure in this link:
http://www.tubecad.com/march99/page2.html

SRPP...they're not even sure what it stands for
http://www.tubecad.com/may2000/
http://www.tubecad.com/articles_2002/SR ... index.html

[jimipage]

Yes, I'd like a schematic and an explanation of that, too.

[krash]

Yeah I'm pretty clear on what he's saying not to use. But I would like to see a real guitar-amp application that accomplishes Aiken's suggested goal of emulating or improving upon EF86 pentode input stage performance using cascode triode arrangement maybe with a 6N1P or whatever dual-triode tube.

I have a hunch there are some commercial amp outfits who are using this technique and keeping it a closely-guarded secret.

[jaysg]

k...I believe the first stage is a self-biased cascode circuit and ken's probably done a good job of it.

http://www.ken-gilbert.com/images/pdf/kenspreamp.pdf

I have gleaned that there are a few tubes meant for this sort of design, and 12a_7's are not. The input to the 65Amps Marquee is probably very much like this.

[mhuss]

I think the issue is that 12A*7s have a fairly low heater-to-cathode maximum (~70 volts IIRC), and the cathode of the top half of the cascode is a bit more than this! The 6DJ8 is great for cascodes, and has been used in a few hi-fi circuits that way, but tends to microphony.

--mark h

[MarkB]

The Firefly has a cascode boost input stage - I'm not sure which bias method it uses. A search of the ax84.com forum should get you plenty of information. As stated above, the 6N1P and 6922 are designed to suit cascode use, and fit the application much better than a 12AX7.

This guy has what looks like a great article on cascode design - I can't vouch for the technical stuff, but it looks solid.

http://www.valvewizard.co.uk/


Here's a useful thread I saved:


--------------------------------------------------------------------------------
From: Matt
Date: 11/4/2002 1:35 PM
Subject: EF86 trial and some Q's (long)


This past weekend I built an EF86 preamp into a "test bed" amp. I started with a 220K plate load resistor, 2.2k cathode resistor bypassed with a 22uF cap, and for the screen grid I used a 1.1M resistor from B+ (same node as the plate, about 340v B+) and a .1uF cap to ground. Suppression grid is connected to the cathode. With these values the screen grid was higher in voltage than the plate (like ~140v screen and ~80v plate IIRC) and of course the tube didn't work. So I changed to a 100K plate load resistor and ended up with ~235v on the plate and ~135v on the screen, cathode is ~2.7v. A couple of questions:

1) I usually see an EF86 with a 220k plate load resistor and Matchless used a 330k plate load resistor, why didn't a 220k work here?

2) Do these numbers that I ended up with sound like a good operating point for this tube (Svetlana EF86)?

3) Since this tube has an internal shield, am I supposed to leave the external tube shield off, or does it even matter?

This amp ended up basically as a plexi-copy circuit except the Bass/Normal channel is the EF86 and since I now had an extra triode I paralleled V1A and V1B for the Bright channel (and adjusted values accordingly). I just used one input and paralleled the control grids of the two channels. I compared it to my homebrew plexi-copy and was THRILLED with the sound. But I can't leave well enough alone. Any help appreciated. Thanks.

Matt



--------------------------------------------------------------------------------
From: Ray Ivers (RAGEray@aol.com)
Date: 11/4/2002 2:36 PM
Subject: Re: EF86 trial and some Q's (long)


Matt,

I would suggest using a voltage-divider for the screen; maybe a 1M from B+ to screen and another (bypassed) 1M from screen to ground would be a good starting point. This can be altered later on, but should give you a low enough Vs to allow you more latitude in Rp values. It looks like you're running a little more than 1 mA Ik right now; I would try 1.5 to 2.5 mA and see if you like the tone any better. Higher currents will obviously drop more voltage across Rp - it's a balancing act.

I don't think I would bother with the shield, myself.

Take a look here here - http://www.duncanamps.co.uk/cgi-bin/tds ... A&DES=EF86 for tube data and a bunch of links to EF86 circuits (you'll have to copy and paste, the link doesn't work right).

Ray Ivers



--------------------------------------------------------------------------------
From: Ray Ivers (RAGEray@aol.com)
Date: 11/4/2002 2:41 PM
Subject: Correct Link


Matt,

Here's the correct link - just type in 'ef86' and you should be good to go.

Ray



--------------------------------------------------------------------------------
From: Matt
Date: 11/4/2002 5:42 PM
Subject: Re: Correct Link


Thanks Ray. Actually the first link you posted worked. I had looked at some Matchless and Vox schematics and just worked from there. I didn't see any use of dividers there and I'm not sure why it didn't occur to me to do that. I'll also try lowering the cathode resistor.

One thing that I found myself doing when playing thru this amp with the new circuit was really hitting the strings hard. I could hit full chords hard and the amp seemed to compress and stay clean. Has a certain clarity to it I like and it LOVES OD pedals. But it also has a 12AT7 long-tail driver with 22k plate loads and about 9mA plate currents so maybe it's that. I'm much too haphazard in my experimentation! But sometimes even a blind squirrel finds a nut.

Thanks.

Matt



--------------------------------------------------------------------------------
From: Morris (mitchellm@gosaints.org)
Date: 11/4/2002 6:25 PM
Subject: Re: EF86 trial and some Q's (long)


Matt,

I've used the exact setup you had initially (220k plate with 1M screen fed from the same B+ node) with an EF86 and it works perfectly. I had around 90V on the plate and 150V on the screen. If yours didn't work, something else was wrong.

Hope this helps.

Morris



--------------------------------------------------------------------------------
From: Matt
Date: 11/4/2002 7:34 PM
Subject: Re: EF86 trial and some Q's (long)


Maybe I touched a solder joint that was bad or something when I swapped plate load resistors. When I first tried it I didn't get any sound, not even hiss. Then again, it wouldn't be the first time I did something completely stupid (like not plugging the cable in, not turning up volume on guitar, etc.). I guess I should go back and try it again. Thanks.

Matt



--------------------------------------------------------------------------------
From: SteveG
Date: 11/10/2002 2:59 PM
Subject: Re: EF86 trial and some Q's (long)


I just finished an AC30/4 clone with the initial values you mentioned, and it works fine.

Boy, can those EF86's be noisey though. I've tried three so far:

Old Mullard - unusable and extremely microphonic.
New 'Edicron' (EI?) - Very microphonic.
Newer Mullard (solid metal screen) - The only one that really worked. Still microphonic, but not unusably so.

Anyone know of a really quiet new production EF86?



--------------------------------------------------------------------------------
From: Mark Lavelle (amps@harmonicappliances.com)
Date: 11/10/2002 4:17 PM
Subject: Re: EF86 trial and some Q's (long)


Try Svetlanas. I've got one working just fine in my Verberator (but it's only a 2 x EL84).



--------------------------------------------------------------------------------
From: Steve A. (steve_ahola@yahoo.com)
Date: 11/10/2002 4:30 PM
Subject: Using a cascode instead of an EF86?


Matt:

I was wondering if anyone has tried replacing an EF86 with a cascoded dual-triode tube, as per this post from Randall Aiken:

http://www.blueguitar.org/new/text/thre ... p/6dj8.txt

In particular whether someone has tried replacing the EF86 in a DC-30 style amp with a cascode...?

--Thanks!

Steve Ahola

P.S. Any explanations about cascodes would be appreciated!



--------------------------------------------------------------------------------
From: Mark C
Date: 11/10/2002 9:58 PM
Subject: Re: Using a cascode instead of an EF86?


TRY it, it is very cool I also built a mic pre with this too.Cascode is a very under used circuit



--------------------------------------------------------------------------------
From: Steve A. (steve_ahola@yahoo.com)
Date: 11/10/2002 10:14 PM
Subject: Looking for a schematic of a cascode


Mark:

After looking through a few pages of Google "hits", I found a reference to the white-tolex bassman head using a cascode on the Bass channel. So that would be the first two stages on the 6G6, 6G6-A and 6G6-B drawings? Maybe I'm missing something but it sure looks like a DC-coupled pair w/ cathode follower to me (using a 220k plate resistor instead of 100k).

Most of the hits from Google were for ss circuits.

Steve Ahola



--------------------------------------------------------------------------------
From: Steve A. (steve_ahola@yahoo.com)
Date: 11/10/2002 10:24 PM
Subject: Re: Looking for a schematic of a cascode


Here's one I just found, but maybe someone can explain it to us...

http://www.triodeel.com/mk3web.gif

Steve Ahola



--------------------------------------------------------------------------------
From: kg (ride5000@ride.ri.net)
Date: 11/10/2002 11:30 PM
Subject: here you go


steve,

here's an old schemo of my preamp...
http://www.ken-gilbert.com/images/pdf/kenspreamp.pdf

another description of the cascode...
http://www.ken-gilbert.com/techstuff/cascode_email.html

another description of one, with jfet bottom and tube top...
http://ken-gilbert.com/techstuff/morepremods.html

have fun!

ken



--------------------------------------------------------------------------------
From: Steve A. (steve_ahola@yahoo.com)
Date: 11/11/2002 12:21 AM
Subject: Re: here you go


ken:

The middle link came up in a Google search for "cascode..." so I had just read through that article... your site looks great! Now that you are registered as "ken-gilbert.com" do you have to pay corporate taxes???

I see that in your preamp you have biased the "upper" stage grid with a 1M resistor to the cathode... I guess that is taking Randall's suggestion one step further? (I.e, not biasing the grid from the B+ rail.) I also see that there is no grid load resistor for that stage... how does that work?

So would you need to change the design if a 6DJ8 was used instead of a 6N1P... or should I just order some 6N1P's?

Is there any way that a 12AX7 could be used for the cascode, like maybe using a split plate load on the top half to keep the gain down a bit? Randall's article mentioned that the cathode to heater voltage of a 12AX7 shouldn't exceed 100v but I've seen 175vdc and up on a 12AX7 used as a cathode follower. I'd like to use 12AX7's because I have so many different varieties and I could pick and choose... but if it's better to use a 6N1P or 6DJ8 I guess I ought to just bite the bullet and order some!

--Thanks!

Steve Ahola



--------------------------------------------------------------------------------
From: kg (ride5000@ride.ri.net)
Date: 11/11/2002 1:50 PM
Subject: Re: here you go


steve,

you could pretty much sub a 6922/6dj8 right into the circuit with very few modifications. you might want to make the rk adjustable so that you can dial in the plate current/plate voltage.

keep in mind one thing about cascodes though--you CANNOT pull the top tube's plate lower in voltage than its grid.

as a result you need to run a high b+ if you expect a high output voltage swing.

the "self-biased" top tube arrangement works by developing its bias voltage across the internal cathode impedance of the tube, approximately 1/gm. i found it less clinical, and more tubey, than a rigidly fixed top tube grid bias.

the fixed top tube grid bias IS easier to adjust, though.

ken



--------------------------------------------------------------------------------
From: Steve A. (steve_ahola@yahoo.com)
Date: 11/10/2002 11:55 PM
Subject: Ok... how is this?


I also found Steve Bench's RIAA preamp using a 12AY7 cascoded into a 6DJ8... let me see if I understand the basic concept here:

The "lower" cascode stage appears to have normal connections to its grid and cathode. However its plate is directly connected to the cathode of the "upper" stage.

Now the "upper" stage plate appears to have a not-too-unusual plate resistor (that would be 220k for a 12AX7?) while the cathode is connected to the plate of the "lower" stage (as mentioned above). The audio signal is taken from the plate of this stage through a coupling cap.

So far so good?

It seems like the grid of the "upper" stage might be the real "bear" in fine-tuning a cascode. The Steve Bench drawing shows the grid of a 6DJ8 connected to a 180k/150k voltage divider going to a B+ node of 185v which puts a voltage of 85v on the grid (the grid also has a 0.1uF cap going to ground).

Randall Aiken's article suggests that the grid be biased with voltage taken off the plate rather than from the power supply node so I'm not sure how that would alter the values of the voltage divider... for this particular circuit you would want to keep roughly 85vdc on the grid- right? Assuming that there is 123vdc on the plate and that you want to keep the grid load at 150k then you would replace the 180k resistor in the voltage divider with a 68k resistor.

So if you were going to use a 6DJ8 for both of the cascoded stages how would this affect the grid load and cathode resistor for the "lower" stage (which is a 12AY7 in the Steve Bench schematic)?

http://my.execpc.com/~n9zes/schem3.html

So many questions, so little time...

--Thanks!

Steve Ahola



--------------------------------------------------------------------------------
From: woodyc (wsjcrane@aol.com)
Date: 11/11/2002 5:43 AM
Subject: Re: Looking for a schematic of a cascode


Steve, If you're interested in how they perform in a guitar amp, I bought some of those boards from Ned a couple years ago and one quickly became a part of my main gigging amp. I just put a triode stage and a tone stack ahead of it and pair of 6550s after. (I'm using a 12AU7 and 12BH7 in the board.) Its tubey-clean, much like a Twin, with less hum and noise.



--------------------------------------------------------------------------------
From: James Ennis (jamesennis@shaw.ca)
Date: 11/10/2002 4:54 PM
Subject: Re: EF86 trial and some Q's (long)


FWIW, when I re-wired the front end of the vi/trem channel of my AC30 be a EF86 channel, I grabbed a Sovtek EF86 to use until I could find a good NOS one. It works great and is still in there, noise free. My friend who has a Matchless went through 3 NOS tubes to find one decent one and still has some rattling. Maybe I just got lucky...Love to hear more about cascode circuits, thanks for bringing that up Steve. Regards,James.



--------------------------------------------------------------------------------
From: Shea (stmoxon@earthlink.net)
Date: 11/11/2002 1:33 AM
Subject: What about a pentode for the 2nd gain stage?


I only hear of pentodes being used for the first gain stage, and I frequently hear about how noisy they are. Has anyone tried one in a second or third gain stage? Maybe then the noise would be less of a problem.

Shea



--------------------------------------------------------------------------------
From: DCH (davechunter@yahoo.co.uk)
Date: 11/11/2002 11:34 AM
Subject: Pentode/triode switching of EF86?


Anyone ever tried a triode switching mod for an EF86 in the first gain stage? I've got an EF86 as a straight pentode in my Matchless D/C30 clone (as designed) and I love it, but a pentode/triode switch might make it a great tube for a one-channel amp but with distinctly different voices: raw, gainy and hot (pentode), or warmer and mellower with less gain - I presume - in triode.

I have been advised by some people to wire the EF86s in my Leak 12+ monoblock hi-fi amps as triodes to make them smoother and less harsh, but haven't tried that yet.

On the other hand, an EF86 as a triode in a guitar amp is using a full 9-pin socket but getting only one triode out of the deal, but if it's switchable it might offer a cool extra voice. I've put triode switching in plenty of output stages and it's certainly simple enough to do. With an EF86, though, would you want a resistor between the screens and plates, or just tie them off straight? Sorry, this turned out to be a long one...

Cheers,

Dave


--------------------------------------------------------------------------------

[krash]

yes I also read that thread... but it raised questions for me rather than providing answers.

The Valvewizard article on cascode in the section Cathode biasing the Upper Triode seems to describe what RA is talking about at least in general terms of topology without verifying the actual values for components that RA was referring to... or how this design compares to a known pentode design like an AC30/4 or something. But at least I guess it's a starting point.

[MarkB]

The design is similar to a pentode in the sense that the gain and the plate output impedence are high. The sound will be different - just like two pentodes are different - but it's more similar to a pentode than a cascaded twin triode. I think that beyond that, you'd just have to build one and listen. There might be cascode-boosted Firefly clips at ax84.com - I'm not sure. You could ask over there.

[drz400]

Yeah I'm pretty clear on what he's saying not to use. But I would like to see a real guitar-amp application that accomplishes Aiken's suggested goal of emulating or improving upon EF86 pentode input stage performance using cascode triode arrangement maybe with a 6N1P or whatever dual-triode tube.

I have a hunch there are some commercial amp outfits who are using this technique and keeping it a closely-guarded secret.

If memory serves me correct I think Kasha did it on a clean channel
Didnt sound too good though!

[drz400]

I think the issue is that 12A*7s have a fairly low heater-to-cathode maximum (~70 volts IIRC), and the cathode of the top half of the cascode is a bit more than this! The 6DJ8 is great for cascodes, and has been used in a few hi-fi circuits that way, but tends to microphony.

--mark h

What if you DC bias the Heater center tap to lets say 70VDC
Good for the hum and heater to cathode voltage no???

[mhuss]

What if you DC bias the Heater center tap to lets say 70VDC
Good for the hum and heater to cathode voltage no???

Sure, if you do it for the one stage that has it cathode "way up in the air." But if you do it across the board, now you're stressing all the other cathodes (which probably have 2-3v on them) as well.

--mark h

[rfgordon]

Here's a cool schematic I found that uses 12AX7 in cascode. I'm thinking of giving this first stage a try in my Franken-amp test bed chassis.

[mhuss]

Actually this is using a 12AU7 (which is much safer). The 12AU7 has a 180v H-K rating.

--mark h

[Elcabong]

Here's a cool schematic I found that uses 12AX7 in cascode. I'm thinking of giving this first stage a try in my Franken-amp test bed chassis.

If you look closely, it uses a 12AU7. I have built something pretty similar, a Cascode Boosted P1, without the reverb. The cascode really push things hard!

But this circuit uses a fixed bias arrangement; it would be interesting to try the "self-bias" version proposed by Aiken.